from __future__ import annotations from datetime import date from typing import Dict, Any, Optional from src.infra.document_analysis.models.company_profile import CompanyProfile from src.infra.document_analysis.normalizers.company_name_normalizer import normalize_company_name from src.infra.document_analysis.normalizers.date_normalizer import parse_date_flexible, months_since def _safe_float(x: Any, default: float = 0.0) -> float: try: return float(x) except Exception: return default def _extract_entity_type_sigla(text: Optional[str]) -> Optional[str]: """Extrai sigla (LDA, SA) de uma descrição ou nome.""" if not text: return None text = text.upper() if "LDA" in text: return "LDA" if "LIMITADA" in text: return "LDA" if "SA" in text or "S.A." in text: return "SA" if "SU, LDA" in text: return "LDA" return None def build_profile_from_alvara(features: Dict[str, Any], entities: Dict[str, Any]) -> CompanyProfile: raw = features or {} company_name = normalize_company_name(raw.get("company_name") or entities.get("company_name")) nuit = raw.get("nuit") or entities.get("nuit") # Tenta inferir entity_type do nome da empresa entity_type = _extract_entity_type_sigla(company_name) # Data de emissão do alvará (não é data de constituição) # Incorporation_date não disponível no alvará incorporation_date = None company_age_months = None issuer = raw.get("issuer") or entities.get("issuer") confidence = _safe_float(raw.get("confidence"), 0.0) # Adiciona campos específicos ao raw (já estão, mas garantimos) raw["alvara_number"] = raw.get("alvara_number") raw["decreto_number"] = raw.get("decreto_number") raw["atividade_principal"] = raw.get("atividade_principal") raw["endereco"] = raw.get("endereco") raw["data_emissao"] = raw.get("data_emissao") raw["codigo_verificacao"] = raw.get("codigo_verificacao") raw["validity_status"] = raw.get("validity_status") return CompanyProfile( source="alvara", company_name=company_name, entity_type=entity_type, nuit=str(nuit) if nuit else None, incorporation_date=incorporation_date, company_age_months=company_age_months, issuer=issuer, confidence=confidence, raw=raw, ) def build_profile_from_certidao(features: Dict[str, Any], entities: Dict[str, Any]) -> CompanyProfile: raw = features or {} company_name = normalize_company_name(raw.get("company_name") or entities.get("company_name")) # entity_type pode vir como sigla (raw["entity_type"]) ou descrição completa entity_type_sigla = raw.get("entity_type") # já é sigla (LDA) no analyzer atual if not entity_type_sigla: entity_type_sigla = _extract_entity_type_sigla(raw.get("entity_type_description") or entities.get("entity_type")) nuit = raw.get("nuit") or entities.get("nuit") # cuidado: na certidão não há NUIT, mas pode vir de entidades erroneamente # Data de constituição (confiável) incorporation_date = raw.get("incorporation_date") if isinstance(incorporation_date, str): incorporation_date = parse_date_flexible(incorporation_date) company_age_months = raw.get("company_age_months") if (company_age_months is None or company_age_months == 0) and isinstance(incorporation_date, date): company_age_months = months_since(incorporation_date) issuer = raw.get("issuer") or entities.get("issuer") confidence = _safe_float(raw.get("confidence"), 0.0) # Campos específicos raw["entity_legal_number"] = raw.get("entity_legal_number") raw["capital_social"] = raw.get("capital_social") raw["capital_realizado"] = raw.get("capital_realizado") raw["beneficiarios_efectivos"] = raw.get("beneficiarios_efectivos") raw["administrator"] = raw.get("administrator") raw["issuance_date"] = raw.get("issuance_date") return CompanyProfile( source="certidao_registo", company_name=company_name, entity_type=entity_type_sigla, nuit=str(nuit) if nuit else None, incorporation_date=incorporation_date if isinstance(incorporation_date, date) else None, company_age_months=company_age_months, issuer=issuer, confidence=confidence, raw=raw, ) def build_profile_from_nuit(features: Dict[str, Any], entities: Dict[str, Any]) -> CompanyProfile: raw = features or {} company_name = normalize_company_name(raw.get("company_name") or entities.get("company_name")) # entity_type pode vir como descrição completa, extrair sigla entity_type_sigla = _extract_entity_type_sigla(raw.get("entity_type") or entities.get("entity_type")) nuit = raw.get("nuit") or entities.get("nuit") # Data de constituição (confiável) incorporation_date = raw.get("incorporation_date") if isinstance(incorporation_date, str): incorporation_date = parse_date_flexible(incorporation_date) company_age_months = raw.get("company_age_months") if company_age_months is None and isinstance(incorporation_date, date): company_age_months = months_since(incorporation_date) issuer = raw.get("issuer") or entities.get("issuer") confidence = _safe_float(raw.get("confidence"), 0.0) # Campos específicos raw["trade_name"] = raw.get("trade_name") raw["incorporation_date_str"] = raw.get("incorporation_date_str") raw["issuer_valid"] = raw.get("issuer_valid") raw["subject_valid"] = raw.get("subject_valid") return CompanyProfile( source="nuit", company_name=company_name, entity_type=entity_type_sigla, nuit=str(nuit) if nuit else None, incorporation_date=incorporation_date if isinstance(incorporation_date, date) else None, company_age_months=company_age_months, issuer=issuer, confidence=confidence, raw=raw, ) def build_profile_from_input(input_data: Dict[str, Any]) -> CompanyProfile: company_name = normalize_company_name(input_data.get("company_name")) entity_type = _extract_entity_type_sigla(input_data.get("entity_type")) nuit = input_data.get("nuit") incorporation_date = input_data.get("incorporation_date") if isinstance(incorporation_date, str): incorporation_date = parse_date_flexible(incorporation_date) company_age_months = input_data.get("months_activity") issuer = "user_input" return CompanyProfile( source="input", company_name=company_name, entity_type=entity_type, nuit=str(nuit) if nuit else None, incorporation_date=incorporation_date if isinstance(incorporation_date, date) else None, company_age_months=company_age_months, issuer=issuer, confidence=1.0, raw=input_data, )